Abstract:
Methods for classifying functions as belonging to a particular software system is disclosed. Various embodiments are also described that use and display the result of the function classification in various ways, regardless of whether the information was generated by any particular of the described methods. One example graph combines function membership data, time interval frame data, and call stack profile data in the form of an Icicle Graph. The Icicle Graph is then distorted into a Sunburst Graph, resulting in an overall Software System Sunburst Icicle Graph.
Abstract:
Profile data collected through sampling is transformed and visually and interactively manipulated and displayed. In one scenario, the profile data is collected through statistical profiling of a program for a video game. The game program profile data is visually and interactively manipulated and displayed to visually present a dynamic behavior of the video game that shows correlation of impacts that the functions of the program have on the video game's performance. This allows the developer to identify inefficient sections of the program for optimization.
Abstract:
The present technology is directed to a non-limiting system and method for generating an abstract code execution timeline from repetitive programs, such as video games and simulations. As a non-limiting example, a statistical sampling profile is first taken of the running program, and then the data is converted into several representations. By selecting representative functions based on sampling frequency and call tree hierarchy and then sorting those functions chronologically and hierarchically, an abstract code execution timeline can be generated that summarizes execution within a single repetitive time period, such as a video frame. This timeline is called “Story Mode” since it tells a story of how the program executes over the course of a frame. This abstract code execution timeline can then be used to facilitate program comprehension, primarily for the purposes of modifying the code to improve execution speed.
Abstract:
The present technology is directed to accurately identifying code execution rhythms from a running program, even when they are occasionally irregular. The present technology is capable of detecting and inferring the rhythms without any prior knowledge or expectations of rates or number of rhythms present in a running program. These code execution rhythms can then be used to further understand and analyze the behavior of a running program, for example, by exposing various rhythms that might have been unknown/unrealized, irregularity of rhythms, or by analyzing the behavior of functions on a per frame basis under different rhythms.
Abstract:
Methods and systems for detecting anomalous function execution in a program, such as a video game or simulation program, are described herein. Certain methods attempt to isolate and score functions that behave in a particular manner that is deemed to be problematic within a repetitive program. Other methods can use the repetitive nature of the program to directly compare and isolate problematic functions.